External References

LLM Provider Documentation

• OpenAI API Reference — Chat Completions, function calling, embeddings
• Anthropic Claude Docs — Claude API, tool use, vision
• Ollama Model Library — Available local models
• LM Studio Docs — Local LLM server setup

Relevant Research Papers

• ReAct: Synergizing Reasoning and Acting in Language Models — Foundational paper for tool-using agents
• AutoGen: Enabling Next-Gen LLM Applications — Multi-agent orchestration
• Constitutional AI — Anthropic's safety alignment approach

Related Tools

• AutoGPT — Another autonomous AI agent
• LangChain — LLM application framework
• Semantic Kernel — Microsoft's AI SDK
• Ollama — Run LLMs locally

Learning Resources

• Learn Prompting — Free guide to prompt engineering
• OpenAI Prompt Engineering Guide
• Anthropic Research Blog

Prompt Engineering Resources

These resources will help you write better tasks and prompts for OpenClaw:

• OpenAI Prompt Engineering Guide — official best practices for GPT models
• Anthropic Prompting Guide — Claude-specific techniques including XML tags and chaining
• PromptingGuide.ai — comprehensive, model-agnostic prompt engineering handbook
• LearnPrompting.org — beginner-friendly interactive course

AI Agent Research Papers

• AgentBench: Evaluating LLMs as Agents (2023) — benchmarks for LLM-based autonomous agents
• ReAct: Synergizing Reasoning and Acting in LLMs — the foundational paper for tool-using agents
• Tree of Thoughts: Deliberate Problem Solving with LLMs — advanced reasoning patterns

LLM Provider Documentation

Developer Tools & Ecosystem

• Ollama — run LLMs locally, the easiest way
• LM Studio — GUI for local models with OpenAI-compatible API
• llama.cpp — low-level inference engine for CPU/GPU
• LangChain — Python framework for LLM-powered apps (complementary to OpenClaw)
• jq — essential tool for processing JSON output from OpenClaw
• n8n — self-hosted workflow automation that can trigger OpenClaw via webhooks

Community & Blog Resources

• Simon Willison's Weblog — essential reading for practical LLM usage and safety
• Latent Space Podcast — deep technical interviews on AI engineering
• Lilian Weng's Blog (OpenAI) — the definitive technical blog on agents, RLHF, and emergent capabilities
• Hugging Face Blog — open-source model releases, fine-tuning guides, benchmarks
• BAIR Blog — academic AI research in accessible format

Benchmarks & Leaderboards

• LMSYS Chatbot Arena — live ELO-based model ranking by crowdsourced human preference
• Open LLM Leaderboard — benchmarks for open-source models
• Artificial Analysis — independent speed, cost, and quality benchmarks for API providers

Security & Safety References

• OWASP LLM Top 10 — the definitive security risk list for LLM applications. Essential reading before production deployment
• NIST AI Risk Management Framework — US government guidelines for responsible AI deployment
• Prompt Injection Attacks and Defences (2023) — academic coverage of the main LLM application attack vector

Tools for Building with LLMs

• LiteLLM — unified Python interface to 100+ LLM APIs (used internally by OpenClaw)
• Microsoft Guidance — structured generation and constrained decoding
• DeepEval — LLM output evaluation and regression testing framework

Related Automation Frameworks

OpenClaw works well alongside these tools — or as a replacement for specific use cases:

Benchmarks & Evaluations

Use these trusted evaluation frameworks to compare models and measure agent quality objectively:

Track the latest model benchmark rankings at LMSYS Chatbot Arena and Open LLM Leaderboard.